The study of cardiovascular disease (CVD) is complex and involves rapidly evolving technologies that enable detection and quantification of subclinical disease over time. Improved analysis of cardiovascular cohort data will result in better identification of epidemiological predictors of cardiovascular disease, improved risk stratification and fewer misleading studies due to inappropriate analysis. The aims of this project are designed to address current methodological problems facing cardiovascular disease researchers. Our goal is to develop novel statistical tools that can help answer new scientific questions using data that has previously been collected by existing NIH-funded studies. We will illustrate our methods with data from two NHLBI-funded prospective cohort studies: The Multi-Ethnic Study of Atherosclerosis (MESA) and The Cardiovascular Health Study (CHS), although the methods and models are generally applicable to many studies. Our grant will address statistical problems across two broad aspects of cardiovascular research where innovative analysis will allow for better use of the information contained in cardiovascular cohort studies. First, we propose new models to handle the methodologic challenge of medication use when studying biomarker associations. Secondly, we develop improved methods for analysis of coronary artery calcium that allow us make better use of the information from existing scans.